On Fri, Feb 16, 2024 at 11:31:01AM +0200, Ilpo Järvinen wrote: > On Thu, 15 Feb 2024, Andy Shevchenko wrote: > > On Thu, Feb 15, 2024 at 06:40:15PM +0200, Ilpo Järvinen wrote: > > > On Thu, 15 Feb 2024, Andy Shevchenko wrote: ... > > > > + /* > > > > + * Nuvoton calls the scratch register 'UART_TOR' (timeout > > > > + * register). Enable it, and set TIOC (timeout interrupt > > > > + * comparator) to be 0x20 for correct operation. > > > > + */ > > > > + serial_port_out(port, UART_NPCM_TOR, UART_NPCM_TOIE | 0x20); > > > > + > > > > + return serial8250_do_startup(port); > > > > > > I know this matches how it is currently done\ > > > > Exactly, I haven't changed the workflow. > > Does it mean you are okay with the change? > > Mostly. Another thing I was let bit unsure if it's okay to move that > serial_port_out() outside of RPM get/put that is inside > serial8250_do_startup(). We have other (actively used AFAIK) drivers which do the same. In any case this driver does not use RPM anyway, and in the future, when Tony finalizes his RPM work those 8250 RPM wrappers should gone. If somebody implements RPM in this driver, it may be done via standard RPM calls. TL;DR: for now it's okay, I am sure. > > > but I wonder if TOIE should not be enabled until ->setup_irq() > > > has been called. -- With Best Regards, Andy Shevchenko
